GPT-6.1 Sol: Release Intelligence, Performance and Price (artificialanalysis.ai)

The release of GPT-6.1 Sol introduces five distinct models, varying significantly in intelligence, performance, and pricing, marking a notable advancement in AI capabilities. The leading model, GPT-6.1 Sol (max), scores an impressive 52 on the Artificial Analysis Intelligence Index, demonstrating enhanced cognitive abilities. In terms of performance, the GPT-6.1 Sol (low) excels with an output speed of 74 tokens per second and remarkably low latency at 1.84 seconds for its first answer token. This model is also the most cost-effective, priced at just $0.13 per task, providing substantial savings compared to its counterparts. This release is significant for the AI/ML community, as it showcases substantial improvements in both speed and efficiency in generating responses, which can be pivotal for applications requiring quick turnarounds, such as real-time customer support or interactive AI systems. The models support both text and image inputs with a generous context window of approximately 1 million tokens, equivalent to 1500 pages of text. The structured evaluation metrics, including the Intelligence Index, equip developers and researchers with detailed insights into the models’ performance and cost-effectiveness, potentially reshaping how AI applications are developed and deployed across industries.